Score: 1

Sharp Minima Can Generalize: A Loss Landscape Perspective On Data

Published: November 6, 2025 | arXiv ID: 2511.04808v1

By: Raymond Fan, Bryce Sandlund, Lin Myat Ko

Potential Business Impact:

More data helps computers learn better from examples.

Business Areas:
Predictive Analytics Artificial Intelligence, Data and Analytics, Software

The volume hypothesis suggests deep learning is effective because it is likely to find flat minima due to their large volumes, and flat minima generalize well. This picture does not explain the role of large datasets in generalization. Measuring minima volumes under varying amounts of training data reveals sharp minima which generalize well exist, but are unlikely to be found due to their small volumes. Increasing data changes the loss landscape, such that previously small generalizing minima become (relatively) large.

Repos / Data Links

Page Count
36 pages

Category
Computer Science:
Machine Learning (CS)